Abstract Ecosystem water-use efficiency (WUE) is a critical indicator to investigate the interaction between the terrestrial ecosystem carbon and water cycles. WUE, estimated from gross primary productivity (GPP) and evapotranspiration (ET) based on remote sensing (RS)-based ecosystem models and algorithms (e.g., MODIS (MODerate resolution Imaging Spectroradiometer), BESS (Breathing Earth System Simulator)), have been used to quantify the spatiotemporal dynamics of WUE and its responses to environmental changes. However, few studies have assessed the ability of RS-based ecosystem models and algorithms on global WUE estimation. In this study, we evaluated 8-day and annual WUE from MODIS and BESS among different sites, land cover types and climate zones using the FLUXNET2015 dataset as reference, and conducted spatial intercomparisons of annual WUE between MODIS, BESS and an upscaled FLUXNET dataset (MTE). The site level evaluation results showed that BESS WUE had better performance than MODIS WUE at both 8-day and annual scales. Among different land cover types and climate zones, MODIS and BESS WUE performed unsatisfactorily, especially for MODIS WUE in open shrublands and savannas and for BESS WUE in closed shrublands. Additionally, both MODIS and BESS WUE performed poorly in the hot semi-arid climate zone. The spatial intercomparisons over 2001-2011 revealed that BESS WUE had similar spatial patterns of annual WUE and linear trends with MTE WUE over the globe, except at the high latitudes. However, the spatiotemporal patterns of MODIS WUE were different from those of MTE and BESS WUE, particularly in the (sub) tropical arid and semi-arid regions. Our evaluations results suggested that coupling carbon and water cycles into RS-based models could improve their performance on global WUE estimation. Moreover, the performance of MODIS and BESS on global WUE estimation should be further improved, especially for their performance on temporal variation and their performance at the (semi) arid areas and the high latitudes.

中文翻译：

---

基于遥感模型对全球用水效率估计的潜力：使用站点级别和放大的涡流协方差数据对生态系统模型 (BESS) 和算法 (MODIS) 进行评估和比对

摘要 生态系统用水效率（WUE）是研究陆地生态系统碳和水循环相互作用的关键指标。WUE 是根据基于遥感 (RS) 的生态系统模型和算法（例如 MODIS（中分辨率成像光谱仪）、BESS（呼吸地球系统模拟器））根据总初级生产力 (GPP) 和蒸散量 (ET) 估算的。用于量化 WUE 的时空动态及其对环境变化的响应。然而，很少有研究评估基于 RS 的生态系统模型和算法对全球 WUE 估计的能力。在本研究中，我们使用 FLUXNET2015 数据集作为参考，评估了 MODIS 和 BESS 在不同地点、土地覆盖类型和气候带之间的 8 天和年度 WUE，并在 MODIS、BESS 和放大的 FLUXNET 数据集 (MTE) 之间进行了年度 WUE 的空间比对。站点级评估结果表明，BESS WUE 在 8 天和年尺度上均优于 MODIS WUE。在不同的土地覆盖类型和气候带中，MODIS 和 BESS WUE 的表现不尽如人意，尤其是在开阔灌木地和稀树草原中的 MODIS WUE 以及在封闭灌木地中的 BESS WUE。此外，MODIS 和 BESS WUE 在炎热的半干旱气候区表现不佳。2001-2011 年的空间比对表明，除高纬度地区外，BESS WUE 与 MTE WUE 在全球具有相似的年度 WUE 空间格局和线性趋势。然而，MODIS WUE 的时空模式与 MTE 和 BESS WUE 的时空模式不同，特别是在（亚）热带干旱和半干旱地区。我们的评估结果表明，将碳循环和水循环耦合到基于 RS 的模型中可以提高其在全球 WUE 估计中的性能。此外，MODIS 和 BESS 在全球 WUE 估计上的性能应进一步提高，特别是它们在时间变化方面的性能以及在（半）干旱地区和高纬度地区的性能。